Dear Allesandro,
In fact, having tried out a few simulations,
I would not recommend using -bootstrap- with -diagt-.
The binomial distribution is thoroughly explotred, and
there are much better ways than bootstrapping of putting
a confidence interval round a percentage.
Several are available in Stata (Wilson Agresti, Jefreys, Exact),
using -ci- or -cii- with much less trouble.
If you do want to, (and I advise against) you need to use options
-bca- and -strata()- with -bootstrap-. The program below
works correctly; (the previous one had true & test confused)
Remove or change the sample -diagti- command to suit your data.
Increase the number of reps to the maximum you can wait for.
Correction to program follows:
*****************************
prog drop _all
prog define mydiagt, rclass
*! mydiagt version 1.0.1 written 12 May 2009 by PT Seed
([email protected])
*! stripped-down version of -diagt- for use with -bootstrap- command.
*! bug in version 1.0 fixed.
* Note: this assumes that values are 0 and 1 only.
version 6.0
* Syntax
syntax varlist(min=2 max=2 numeric) [if]
tempvar touse
mark `touse' `if'
tokenize "`varlist'"
local true `1'
local test `2'
su `test' if `true' == 1 & `touse', meanonly
return scalar sens = r(mean)
su `test' if `true' == 0 & `touse', meanonly
return scalar spec = (1-r(mean))
end mydiagt
clear
diagti 4 2 7 37
qui bootstrap sens=r(sens) spec=r(spec), strata(true) reps(10000) bca:
mydiagt true test
estat bootstrap, all
--
Paul T Seed MSc CStat CSci, Senior Lecturer in Medical Statistics,
tel (+44) (0) 20 7188 3642, fax (+44) (0) 20 7620 1227
Wednesdays: (+44) (0) 20 7848 4208
[email protected], [email protected]
King's College London, Division of Reproduction and Endocrinology
St Thomas' Hospital, Westminster Bridge Road, London SE1 7EH
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
